Learning to Solve Multiple-TSP With Time Window and Rejections via Deep Reinforcement Learning

被引:26
作者
Zhang, Rongkai [1 ]
Zhang, Cong [2 ]
Cao, Zhiguang [3 ]
Song, Wen [4 ]
Tan, Puay Siew [3 ]
Zhang, Jie [2 ]
Wen, Bihan [1 ]
Dauwels, Justin [5 ]
机构
[1] Nanyang Technol Univ, Sch Elect & Elect Engn, Singapore 639798, Singapore
[2] Nanyang Technol Univ, Sch Comp Sci & Engn, Singapore 639798, Singapore
[3] ASTAR, Singapore Inst Mfg Technol SIMTech, Singapore 138632, Singapore
[4] Shandong Univ, Inst Marine Sci & Technol, Qingdao 266237, Peoples R China
[5] Delft Univ Technol, Fac EEMCS, Dept Microelect, NL-2628 CD Delft, Netherlands
基金
中国国家自然科学基金;
关键词
Task analysis; Costs; Routing; Reinforcement learning; Time factors; Training data; Market research; Travelling salesman problem; graph neural network; deep reinforcement learning; ALGORITHM;
D O I
10.1109/TITS.2022.3207011
中图分类号
TU [建筑科学];
学科分类号
0813 ;
摘要
We propose a manager-worker framework (the implementation of our model is publically available at: https://github.com/zcaicaros/manager-worker-mtsptwr) based on deep reinforcement learning to tackle a hard yet nontrivial variant of Travelling Salesman Problem (TSP), i.e. multiple-vehicle TSP with time window and rejections (mTSPTWR), where customers who cannot be served before the deadline are subject to rejections. Particularly, in the proposed framework, a manager agent learns to divide mTSPTWR into sub-routing tasks by assigning customers to each vehicle via a Graph Isomorphism Network (GIN) based policy network. A worker agent learns to solve sub-routing tasks by minimizing the cost in terms of both tour length and rejection rate for each vehicle, the maximum of which is then fed back to the manager agent to learn better assignments. Experimental results demonstrate that the proposed framework outperforms strong baselines in terms of higher solution quality and shorter computation time. More importantly, the trained agents also achieve competitive performance for solving unseen larger instances.
引用
收藏
页码:1325 / 1336
页数:12
相关论文
共 36 条
[1]   An integrated production and transportation scheduling problem with order acceptance and resource allocation decisions [J].
Aminzadegan, Sajede ;
Tamannaei, Mohammad ;
Fazeli, Majid .
APPLIED SOFT COMPUTING, 2021, 112
[2]   AN EXACT ALGORITHM FOR THE TIME-CONSTRAINED TRAVELING SALESMAN PROBLEM [J].
BAKER, EK .
OPERATIONS RESEARCH, 1983, 31 (05) :938-945
[3]  
Cappart Q, 2021, AAAI CONF ARTIF INTE, V35, P3677
[4]  
Chen MX, 2020, Arxiv, DOI [arXiv:2005.09330, 10.48555/abs-2005.09330, DOI 10.48555/ABS-2005.09330]
[5]   Learning Heuristics for the TSP by Policy Gradient [J].
Deudon, Michel ;
Cournut, Pierre ;
Lacoste, Alexandre ;
Adulyasak, Yossiri ;
Rousseau, Louis-Martin .
INTEGRATION OF CONSTRAINT PROGRAMMING, ARTIFICIAL INTELLIGENCE, AND OPERATIONS RESEARCH, CPAIOR 2018, 2018, 10848 :170-181
[6]  
Fu Z.-H., 2020, arXiv
[7]  
Fu ZH, 2021, AAAI CONF ARTIF INTE, V35, P7474
[8]  
Heris M. K., 2019, YPEA YARPIZ EVOLUTIO
[9]   A survey of dial-a-ride problems: Literature review and recent developments [J].
Ho, Sin C. ;
Szeto, W. Y. ;
Kuo, Yong-Hong ;
Leung, Janny M. Y. ;
Petering, Matthew ;
Tou, Terence W. H. .
TRANSPORTATION RESEARCH PART B-METHODOLOGICAL, 2018, 111 :395-421
[10]   A reinforcement learning approach for optimizing multiple traveling salesman problems over graphs [J].
Hu, Yujiao ;
Yao, Yuan ;
Lee, Wee Sun .
KNOWLEDGE-BASED SYSTEMS, 2020, 204